ORIGINAL ARTICLES

BIOPRESERVATION AND BIOBANKING Volume 12, Number 4, 2014 ª Mary Ann Liebert, Inc. DOI: 10.1089/bio.2014.0001

Development of a Pilot Project on Data Sharing among Partners of the Italian Hub of Population Biobanks (HIBP): Association between Lipid Profile and Socio-Demographic Variables Mariarosaria Napolitano,1 Filippo Santoro,2 Maria Puopolo,2 Chiara Donfancesco,3 Lucia Galluzzo,3 Alessandro De Grandi,4 Elisa Cevenini,5 Amalia De Curtis,6 Federica Sevini,5 Luigi Palmieri,3 Deborah Mascalzon,4 Paolo Roazzi,7 Emanuele Scafato,3 Peter Pramstaller,4 Licia Iacoviello,6 Maria B. Donati,6 Simona Giampaoli,3 Claudio Franceschi,5 Filippo Belardelli,1 and Elena Bravo1

The Italian Hub of Population Biobanks (HIBP) includes both ongoing and completed studies that are heterogeneous in both their purpose and in the specimens collected. The heterogeneity in starting conditions makes sharing study data very difficult because of technical, ethical, and collection rights issues that hamper collaboration and synergy. With the aim of overcoming these difficulties and establishing the ‘‘proof-of-concept’’ that sharing studies is achievable among Italian collections, a data-sharing pilot project has been agreed to by HIBP members. Participants agreed to the general methodology and signed a shared Data Transfer Agreement. The biobanks involved were: EURAC (Micros study), CIG (GEHA project), CNESPS (FINE, MATISS, MONICA, OEC1998, ITR (Italian Twin Register), and IPREA studies, and MOLIBANK (Moli-Sani project). Biobank data were uploaded into a common database using a dedicated informatics infrastructure. Demographic data, and anthropometric and hematochemical parameters were shared for each record. Each biobank uploaded into the common database a dataset with a minimum of 1000 subjects, for a total of 5071 records. After a harmonization process, the final dataset included 3882 records. Subjects were grouped into three main geographic areas of Italy (North, Center, and South) and separate analyses were performed for men and women. The 3882 records were analyzed through multivariate logistic regression analysis. Results were expressed as odds ratios with 95% confidence interval. Results show several geographical differences in the lipidemic pattern, mostly regarding cholesterol-HDL, which represents a strong basis for further, deeper sample-based studies. This HIBP pilot study aimed to prove the feasibility of such collaborations and it provides a methodological prototype for future studies based on the participation in the partnership of well-established quality collections.

Introduction

B

ioresources play a critical role in the advancement of biotechnology and disease-related research.1 In particular, collections of biological samples from populationbased studies, accompanied by related phenotype, genotype, clinical, and research data, represent a pivotal tool for epidemiological and genetic studies such as those identifying genes responsible for complex diseases and risk factors, and for planning effective preventive actions for public health. The

different research communities of population biobanks share general overlapping concepts and knowledge; however, they differ significantly in sampling approaches, ethical accomplishments, vocabularies, and workflows. Thus, the definition of ‘‘suitable for use’’ for data varies considerably, dependent on the specific research questions addressed by the individual collections. The heterogeneity of sample and data management makes sharing data very difficult due to technical, ethical, and collection rights issues which hamper collaboration

Departments of 1Haematology, Oncology and Molecular Medicine, 2Cell Biology and Neurosciences, 3National Center for Epidemiology, Surveillance and Health Promotion, and 7Information Technology, Istituto Superiore di Sanita`, Rome, Italy. 4 Center for Biomedicine, EURAC Research, Bolzano, Italy. 5 C.I.G Interdipartimental Center ‘‘L.Galvani’’, and DIMES—Department of Experimental, Diagnostic and Specialty Medicine, University of Bologna, Bologna, Italy. 6 Department of Epidemiology and Prevention IRCCS Istituto Neurologico Mediterraneo Neuromed, Pozzilli, Italy.

225

226

NAPOLITANO ET AL.

and synergy.2 Thus, there is much to be gained from coordination, combination, and integration. With the objective of overcoming these difficulties, the Italian Hub of Population Biobanks (HIBP), a network of population collections including both on-going and completed studies, was established in 2011.3 Although many collections and population biobanks are already part of European and International projects and are listed in the catalogue of Public Population Project in Genomics (p3g),4 the number of collaborative projects based on existing collections could be increased considerably. The aim of this study was to show the opportunities provided by the HIBP to share data of different population-based studies. Within the HIBP, biobanks that use similar approaches and methodologies or different solutions and operation workflows may coexist. The HIBP aims to encourage collaboration between these different biobanks, stimulate integration of these infrastructures, and increase the synergy between the study outcomes and Regional and National health plans. To foster interoperability and establish the ‘‘proof-of-concept’’ that ‘‘sharing’’ existing data may provide supplementary epidemiological information, the HIBP partners agreed to conduct a pilot project. The priority objective was to establish the feasibility of constructing a shared approach to the conduct of this type of project. A second objective was analysis of the data collected from different projects to find statistical associations. In addition to the ability to build such a study, the co-authors envisage that the reliability of the approach could be evaluated by comparing the associations found with current knowledge in the field.

The specific topic selected for the HIBP pilot was assessment of the associations between lipid profile data (hematochemical parameters) and the geographical area of residence, taking into account the possible effects of demographic and socio-cultural factors. Previous work has shown that the lipidemic parameters affecting cardiovascular risk are influenced by geographical gradient.5,6 One of the advantages of integrating data from different population studies is the use of regional data to cover the entire national territory.

Material and Methods The aims and limits of the pilot project were defined during extensive preliminary discussion among the partners, and each agreed to the methodologies to be implemented. This preliminary process required extensive work, as the group needed to achieve a degree of personal knowledge and reciprocal trust. Most of this preparatory work was done during the first phase of construction of the network and the complete publication of collection information on the network website.3 The process was facilitated by a small nucleus of participants who agreed to implement a pilot project. Approximately 1 year after the creation of the HIBP, four meetings were dedicated to defining the terms and text of the Data Transfer Agreement (DTA). The DTA established collaboration rules, data definitions, modalities for sharing and analysis of data, and the time frame of the data to be included. The main features of the pilot project are reported in Table 1.

Table 1. Features of the Pilot Project Biobank

Specific Collection

E U R A C

Micros1

C I G

GEHA

2

C N E S P S

FINE3 MATISS4 MONICA5 OEC19986 ITR7 IPREA8

M O L I B A N K

Moli-Sani9

Terms of Data Transfer Agreement

Steps - Identification of scientific target - Definition of data to share; data record - Verification of the ethical issues and informed consent harmonization - Drafting of Data Transfer Agreement (DTA) - Setup of dataset - File data upload - Data harmonization - Data statistical analysis - Data dissemination

-

Parties Agreement purpose Inclusion criteria for parties Number of records for each biobank Record upload and data management Results sharing Coordinator and parties Ethics policy Personal data policy Informed consent policy Results dissemination and authorship Life-span of the collaboration

Data To Be Shared -

Age at sampling Gender Birthplace Place of residence Weight Height Education Total cholesterol Cholesterol-(LDL) Cholesterol-(HDL) Triglycerides

1 Micros: Study of microisolates in South Tyrol (see Reference 8); 2GEHA: Genetics of Healthy Aging (see Reference 9); 3FINE: Finland Italy the Netherlands Elderly- Italian Cohorts (see Reference 10); 4MATISS: Malattie Aterosclerotiche Istituto Superiore di Sanita’ (see Reference 11); 5MONICA: Area Latina-Monitoring Cardiovascular Disease (see Reference 12); 6OEC 1998: Osservatorio Epidemiologico Cardiovascolare (see Reference 13). 7 ITR: Italian Twin Register (see Reference 14); 8IPREA: Italian Project on the Epidemiology of Alzheimer’s Disease (see Reference 15); 9 Moli-Sani: Moli-Sani project (see Reference 16).

HIBP: DATA SHARING PILOT PROJECT

227

The characteristics of the pilot project were assessed to establish that they were compliant with ethics policies and with the informed consents that had been collected for various studies involved in the HIBP. A basic tool that allowed comparison of consent forms was developed; it enabled identification of areas of work that complied with the original consent form information and those that did not.7 On the basis of the DTA, four categories of data—general data, demographic data, physical parameters and hematochemical parameters—were selected for use (Table 2). Almost all parameters were coded and free text fields were reduced to a minimum. A website was created to allow secure submission of the data. A hypertext transfer protocol over secure socket layer (HTTPS) was adopted. Different degrees of computerization in biobanks were taken into account by accepting three types of file formats: CSV (Comma-Separated Values), XLS/XSLX (Microsoft Excel), and XML (eXtensible Markup Language). A notification message generated by the website at the end of data upload guaranteed the data transfer and receipt. Each partner participating in the pilot project uploaded data for their specific collection: EURAC (Micros-study)8, CIG (GEHA-project)9, CNESPS (FINE,10 MATISS,11 MONICA,12 OEC1998,13 Italian Twin Register,14 and IPREA15 studies), MOLIBANK (Moli-Sani project)16. CNESPS contributed six different collections; since each of these studies had been performed independently the data had not been harmonized. Each record referred to a single subject. Additional information on these studies is available on the network website.3 Harmonization of the final dataset for statistical analysis required considerable work and was independently performed by three people (Table 3). Units of measurement and spelling were made uniform, and the consistency of each record was carefully checked to verify coherence with the

required data. The following exclusion criteria were adopted for the statistical analysis: lack of certainty on the collection of fasting blood, age £ 25 years, residence abroad, and missing data in one of the following fields: place of residence,17 education,18 or body mass index (BMI).19 Moreover, underweight subjects (a pathological condition) were excluded to avoid biases in the reference group of BMI. Several different aspects had to be considered for harmonizing the education data18 and hematochemical parameters.20–22 For the education data (Table 4), the original response recorded by the biobank has been expressed in terms of years of scholarship which relates to the Italian educational system.18 Education was analyzed on a continuous scale on the ordinal harmonized classes (see Table 4) by assuming increments of one unit between subsequent classes. For statistical analysis, two risk categories were determined for each hematochemical parameter, depending on the risk classification (Table 5). Association between ‘‘risk’’ category and place of residence, education, and BMI was estimated by logistic regression analyses, stratified by age range (26–40 years; 41–60 years; 61–79 years; ‡ 80 years) and gender, as per Table 3. Residence in North Italy and normal-overweight BMI (see Table 2) were chosen as reference categories in this analysis. Univariate and multivariate analyses were carried out to obtain crude and adjusted odds ratios (ORs) with 95% confidence interval (95% CI). All statistical analyses were carried out with STATA X11.

Results Preliminary work enabled this pilot project to be set so that it was compliant with the informed consent forms collected by each individual collection.7 The main features

Table 2. Harmonization of Pilot Project Data Data Category

Requested data

General data

Date of sampling Biobank/collection

Demographic data

Age at sampling Gender Birthplace Place of residence (Italian Regions and Provinces) Education

Parameter

Derived parameter No harmonization No harmonization

No harmonization No harmonization No harmonization Categories17 Permanent address - North (only Italian Regions - Center were considered) - South See Table 4

Physical parameters

Weight Height

Body mass index (BMI)

Hematochemical parameters

Total-cholesterol LDL HDL Triglycerides

1. Data on fresh and frozen biological matrix were considered equivalent20,21 2. All values were converted to mg/dl20,21 3. Data on plasma-EDTA were harmonized to those sampled in serum20,21

Categories19 Under weight: ‡ 0, < 18.5 Normal weight: ‡ 18.5, < 25 Over weight: ‡ 25, < 30 Obese: ‡ 30

Interval of reference. (See Table 5)

Before harmonization

Men

Women

Men

Women

0 0 0 310 310

26–40 41–60 61–79 ‡ 80 Total 0 297 297

0 0 0 569 569

26–40 41–60 61–79 ‡ 80 Total

4–25 ‡ 26 Total

0 0 0 0 0 375 375

4–12 13–25 26–40 41–60 61–79 ‡ 80 Total

0 100 100

0 0 0 100 100

0 0 0 100 100

0 0 0 0 0 100 100

0 0 0 0 0 100 100

0 3 3

59 254 168 18 499

48 299 140 11 498

0 0 59 254 169 18 500

0 0 48 300 141 11 500

0 0 0 0 0 801 801

4–12 13–25 26–40 41–60 61–79 ‡ 80 Total

0 100 100

11.8 50.9 33.7 3.6 100

9.6 60.0 28.1 2.2 100

0 0 11.8 50.8 33.8 3.6 100

0 0 9.6 60.0 28.2 2.2 100

%

n

n

%

Moli-Sani

GEHA

Age ranges (years)

172 316 488

164 138 88 4 394

184 180 93 11 468

11 59 214 182 119 5 590

10 92 245 244 153 16 760

n

35.2 64.8 100

41.6 35.0 22.3 1.0 100

39.3 38.5 19.9 2.4 100

1.9 10 36.3 32.9 20.2 0.9 100

1.3 12.1 32.2 32.1 20.1 2.1 100

%

Micros

0 8 8

0 0 211 69 280

0 0 1947 77 271

0 0 0 0 214 70 284

0 0 0 0 197 78 275

n

0 100 100

0 0 75.4 24.6 100

0 0 71.6 28.4 100

0 0 0 0 75.4 24.6 100

0 0 0 0 71.6 28.4 100

%

IPREA

0 1 1

33 33 33 0 99

33 33 34 0 100

0 0 33 33 34 0 100

0 0 33 33 34 0 100

n

0 100 100

33.3 33.3 33.3 0 100

33.0 33.0 34.0 0 100

0 0 33.0 33.0 34.0 0 100

0 0 33.0 33.0 34.0 0 100

%

MATISS n

0 3 3

0 49 48 0 97

0 50 50 0 100

0 0 0 50 50 0 100

0 100 100

0 50.5 49.5 0 100

0 50.0 50.0 0 100

0 0 0 50.0 50.0 0 100

0 0 0 50.0 50.0 0 100

%

MONICA

0 0 0 50 50 0 100

Biobank/collection

0 201 201

0 0 0 0 0

0 0 0 0 0

0 0 0 0 0 201 201

0 0 0 0 0 0 0

n

0 100 100

0 0 0 0 0

0 0 0 0 0

0 0 0 0 0 100 100

0 0 0 0 0 0 0

%

FINE

0 3 3

33 33 34 0 100

32 31 34 0 97

0 0 33 33 34 0 100

0 0 33 33 34 0 100

n

0 100 100

33.0 33.0 34.0 0 100

33.0 32.0 35.0 0 100

0 0 33.0 33.0 34.0 0 100

0 0 33.0 33.0 34.0 0 100

%

OEC1998

Table 3. Effect of Harmonization Process on Number of Shared Records in Pilot Project

Harmonized records have been used for statistical analysis.

Excluded records

After harmonization

228

20 165 185

0 0 0 0 0

0 0 0 0 0

0 6 20 16 24 0 66

0 14 49 33 23 0 119

n

%

10.8 89.2 100

0 0 0 0 0

0 0 0 0 0

0 9.1 30.3 24.2 36.4 0 100

0 11.8 41.2 27.7 19.3 0 100

ITR

192 997 1189

289 507 582 401 1779

297 593 545 668 2103

11 65 359 568 644 669 2316

10 106 408 693 632 906 2755

n

%

16.1 83.9 100

16.3 28.5 32.7 22.5 100

14.1 28.2 25.9 31.8 100

0.5 2.8 15.5 24.5 27.9 28.9 100

0.4 3.9 14.8 25.2 22.9 32.9 100

Total

229

3

4

3

4

1–4

2

Harmonization (years of studies)

1

Any certificate, illiterate or literate

Any certificate

Obtained Italian certificate

OEC98 MONICA MATISS

2

Any certificate

1

Any or elementary education

Harmonization (Italian certificate)

Number of years of education

FINE

Obtained Italian certificate

Moli-Sani

Obtained Italian certificate

What is your highest school education?

Micros

IPREA

Educational data required by questionnaire How many years did you attend school?

GEHA

Educational categorization by dataset 6

7

6

7

5–7

Elementary school Italian certificate

Elementary school Italian certificate

5

Elementary school Italian certificate

Elementary school Italian certificate

Elementary school

5

Minimum five years of education and/or elementary school Italian certificate 9

10

9

8–10

Middle school Italian certificate

Middle school Italian certificate

8

Middle school Italian certificate

Middle school Italian certificate

Secondary school

8

Minimum eight years of education and/or middle school Italian certificate

14

High school

11–15

High school Italian certificate

High school Italian certificate

12

Vocational school

15

High school

High school Italian certificate

Vocational School

14

Three-five years after middle school and/or high school Italian certificate 11 12 13

Table 4. Harmonization of Education Data

*ITR lacked education data and the records were excluded from statistical analysis in accord with the exclusion criteria (see Table 3).

Collection

20

‡ 16

Degree

Degree

18

19

20

2nd level degree

19

1st level degree

University

18

2nd level degree and postgraduate studies

17

1st level degree

16

Graduate and postgraduate Italian certificate

230

NAPOLITANO ET AL.

Table 5. Harmonization of Hematochemical Parameters Hematochemical parameter Total cholesterol

LDL

HDL

Triglycerides

Reference interval (mg/dL)

Classification*

Category for statistical analysis

£ 199 200–239 ‡ 240

Desirable Borderline High

Desirable + Borderline

£ 100 101–129 130–159 160–189 ‡ 190

Very Desirable Desirable Borderline High Very High

Very desirable + Desirable + Borderline

£ 40 41–59 ‡ 60 £ 150 151–199 20–499 ‡ 500

Low Normal High Normal Borderline High Very High

High

High + Very high Low + Normal High Normal + Borderline High + Very high

*According to Reference 22.

of the pilot project, as reported in Table 1, include the terms addressed in the DTA. The DTA was signed by the principal investigators of the studies used in the pilot project. Each biobank uploaded data for a minimum of 1000 subjects, making a total of 5071 records (Table 3, before harmonization). All uploaded records included the age parameter. The first two age ranges (4–12 and 13–25 years) were excluded from statistical analysis because of the low number of subjects and the uneven distribution across participating biobanks. In fact, only two (Micros and ITR) out of nine collections participating in the study collected samples in the age range of 4–25 years. After the harmonization process and the adoption of the exclusion criteria reported in the Materials and Methods section, the statistical file included 3882 records (Table 3). Thus, 23% of the records were excluded, of which 16% was due to the elimination of the 4–12 and 13–25 age ranges and 84% was due to the other exclusion criteria. The summary of the excluded records for each collection is reported in Table 3. Results of multivariate logistic regression analysis, stratified by age and gender, expressed as adjusted ORs and respective confidence intervals, are reported in Table 6. In this table, statistically significant ORs are marked in bold. In women (Table 6B), the results of multivariate logistic regression analysis showed a consistent, significant inverse association (OR < 1) between level of education and each of the considered hematochemical parameters in the age range 41–60. As to the relationship between gender and the significant associations shown in Table 6, the comparison between panel A and B shows that, of the 17 significant associations in men and 22 in women, only 9 are common to both genders. Six of these concern HDL, and with the exclusion of subjects aged over 80 years, high HDL levels are associated with residence in the Center of Italy and with high BMI levels (just missing statistical significance in men in the age range 61–79 years). In subjects of the older age range (over 80 years), high HDL levels are associated with residence in the South of Italy in both men and women. The same positive association is also found in men over 41 years and in women in the age range 26–40 years.

Table 6 also shows that in women, higher levels of LDL are significantly associated with residence in Center Italy with the exclusion of the age range 61–79 years, that only approximate to the statistical significance. In men, the same type of relationship is observed in the age ranges 41–60 and 61–79 years. Interestingly, women and men in the 61–79 age range and living in the South present a decreased risk (OR < 1) of having both high LDL and high total cholesterol levels. Limited to total cholesterol and to women, the same association is also observed in the 26–40 age range. A description of each association found by multivariate logistic regression analysis is reported in Table 6, and will not be discussed in detail. However, it is interesting to note the trend between triglyceride levels and BMI in men. The results indicate a strong positive correlation between increased triglyceride levels and BMI between 26 and 60 years. This association tends to decrease with increasing age (no statistically significant relationship was found between 61 and 79 years); in subjects over 80 years, BMI is negatively correlated with triglycerides.

Discussion Population biobanks require expensive investments23 and in Italy, as elsewhere, policy makers are interested in the potential of biobanks to use their tools for national health plans and biotechnological innovation.24–26 However, it is recognized that due to the lack of harmonized measures, few of the large-scale population cohorts can be used to take full advantage of the rich array of behavioral and other diseaserelated data that accompany many studies, as well as the environmental exposures that are potentially available.27 Thus, it is not surprising that the project for the implementation of the HIBP, supported by the National Center for Disease Prevention and Control (Ccm), included the initiation of a pilot project that aimed to demonstrate that synergy among alreadycollected and available data can be used to address scientific questions on preventive and/or predictive medicine. The overall objective of building a first ‘‘proof-of-concept’’ pilot study (i.e., using pooled data from different cohorts with

231

BMI

Normaloverweight Obese

Normaloverweight Obese

South

North Center

Education

Place of residence

(B)

BMI

South

North Center

Education

Place of residence

(A)

0.79 (0.47–1.33)

0.77 (0.43–1.40)

1.09 (0.86–1.39) 1 3.06 (1.43–6.56) 0.48 (0.26–0.88) 1 1.50 (0.54–4.21)

0.80 (0.53–1.20) 1 1.41 (0.4–4.99) 1.04 (0.43–2.58) 1

‡ 80 (n = 584)

0.79 (0.66–0.96) 1 1.44 (0.85–2.44) 0.83 (0.55–1.24) 1

1.03 (0.68–1.55)

0.78 (0.26–2.34)

0.80 (0.56–1.15)

1.08 (0.90–1.30) 1 1.06 (0.65–1.71) 0.47 (0.30–0.72) 1

61–79 (n = 545)

1.37 (0.83–2.24)

1.05 (0.86–1.29) 1 1.34 (0.8–2.26) 1.38 (0.91–2.10) 1

‡ 80 (n = 584)

1.34 (0.44–4.14)

0.61 (0.36–1.05) 1 2.76 (1.15–6.65) 0.32 (0.07–1.43) 1

26–40 (n = 293)

1.24 (0.79–1.96)

0.62 (0.50–0.77) 1 2.69 (1.34–5.37) 0.65 (0.42–1.00) 1

41–60 (n = 541)

1.0 (0.63–1.59)

1.07 (0.85–1.35) 1 2.04 (0.98–4.24) 0.41 (0.24–0.69) 1

61–79 (n = 374)

1.79 (1.01–3.16)

0.85 (0.65–1.11) 1 2.26 (1.17–4.34) 2.22 (1.3–3.78) 1

‡ 80 (n = 518)

Age ranges

4.42 (2.11–9.25)

0.91 (0.72–1.14) 1 2.43 (1.17–5.04) 0.87 (0.54–1.40) 1

61–79 (n = 545)

Age ranges

41–60 (n = 591)

0.77 (0.49–1.22) 1 1.80 (0.80– 4.05) 0.28 (0.08–0.95) 1

26–40 (n = 294)

1.08 (0.43–2.72)

0.60 (0.39–0.93) 1 1.71 (0.81–3.61) 0.79 (0.38–1.62) 1

41–60 (n = 591)

LDL (high + very high) vs ( very desirable + desirable + borderline)

0.86 (0.53–1.40)

0.76 (0.47–1.22)

2.94 (1.41–6.14)

0.95 (0.69–1.29) 1 1.0 (0.39–2.59) 0.57 (0.27–1.20) 1

26–40 (n = 294)

Total cholesterol (high) vs (desirable + borderline)

1.16 (0.96–1.41) 1 1.34 (0.77–2.31) 0.52 (0.31–0.88) 1

0.93 (0.75–1.16) 1 1.31 (0.74–2.31) 0.88 (0.55–1.39) 1

‡ 80 (n = 377)

Age ranges

61–79 (n = 582)

Age ranges

41–60 (n = 502)

0.79 (0.52–1.20) 1 1.55 (0.74–3.27) 0.87 (0.44–1.72) 1

26–40 (n = 286)

LDL (high + very high) vs (very desirable + desirable + borderline)

Total cholesterol (high) vs (desirable + borderline)

5.59 (2.47–12.68)

0.62 (0.43–0.88) 1 3.22 (1.58–6.57) 3.46 (1.88–6.37) 1

26–40 (n = 294)

2.86 (1.50–5.46)

1.16 (0.91–1.48) 1 2.48 (1.07–6.31) 1.88 (1.18–3.00) 1 1.60 (0.96–2.69)

0.98 (0.79–1.21) 1 2.28 (1.06–5.22) 2.05 (1.26–3.33) 1

61–79 (n = 407)

Age ranges 41–60 (n = 453)

2.69 (1.79–4.05)

0.72 (0.60–0.86) 1 3.95 (1.87–8.35) 0.97 (0.67–1.40) 1

41–60 (n = 541)

2.64 (1.73–4.03)

0.85 (0.69–1.05) 1 2.99 (1.39–6.41) 1.44 (0.90–2.29) 1

61–79 (n = 374)

Age ranges

HDL (low + normal) vs (high)

3.12 (1.06–9.15)

0.76 (0.50–1.14) 1 9.49 (2.21–40.7) 1.37 (0.74–2.55) 1

26–40 (n = 286)

HDL (low + normal) vs (high)

Table 6. Adjusted Odds Ratio (OR) for Men (A) and Women (B)

1.60 (0.96–2.66)

0.87 (0.72–1.06) 1 1.35 (0.81–2.26) 2.69 (1.81–4.01) 1

‡ 80 (n = 523)

1.71 (0.83–3.52)

0.99 (0.81–1.21) 1 1.12 (0.53–2.37) 1.82 (1.10–3.00) 1

‡ 80 (n = 326)

1.95 (0.40–9.45)

0.69 (0.29–1.63) 1 1.52 (0.30–7.63) 1.0 (0.20–4.94) 1

26–40 (n = 294)

2.39 (1.47–3.88)

0.90 (0.70–1.15) 1 1.12 (0.46–2.70) 1.37 (0.83–2.27) 1

1.14 (0.67–1.94)

0.93 (0.74–1.17) 1 0.52 (0.22–1.21) 0.94 (0.54–1.62) 1

61–79 (n = 534)

Age ranges 41–60 (n = 453)

1.62 (0.79–3.33)

0.64 (0.45–0.92) 1 2.84 (0.79–10.20) 2.41 (1.03–5.68) 1

41–60 (n = 541)

1.11 (0.65–1.91)

0.79 (0.59–1.07) 1 1.89 (0.88–4.05) 1.0 (0.52–1.94) 1

61–79 (n = 495)

Age ranges

Triglycerides (high + very high) vs (normal + borderline)

2.37 (1.02–5.51)

0.94 (0.57–1.55) 1 1.38 (0.55–3.49) 1.43 (0.66–3.07) 1

26–40 (n = 286)

0.83 (0.36–1.89)

0.78 (0.54–1.13) 1 1.27 (0.50–3.20) 2.50 (1.26–4.93) 1

‡ 80 (n = 585)

0.46 (0.11–2.00)

1.35 (1.01–1.82) 1 1.05 (0.35–3.17) 0.56 (0.23–1.37) 1

‡ 80 (n = 377)

Triglycerides (high + very high) vs (normal + borderline)

232

available tools and methods to provide important information for public health) has been achieved. With respect to the specific topic addressed in this study, we found that results of multivariate logistic regression analysis reflect some well-known relationships between lipid patterns. Thus, the consistency of the associations found with current knowledge supports the validity of the method developed and of the harmonization process, indicating in addition, the value of pooled data. The HIBP pilot study also shows some new interesting associations. The results provide some evidence that it would be worthwhile to validate these findings by further experimentation. Other studies have reported that education level has a positive effect on some markers related to cardiovascular risk and lipid profile.28,29 In Italy, before the sixties, education levels were very poor, especially for women.17 Most of the people now aged between 40 and 61 years had the opportunity to access a higher level of education and general knowledge. Thus, the consistent association found in our study in women aged 41–60 years between education and a less atherogenic lipid pattern (lower total-cholesterol and LDL, higher HDL and lower triglycerides), allows the speculation that higher education increases awareness of the importance of lifestyle on health and aesthetic appearance. Moreover, the finding that this association is not present in younger and more educated women (26–40 age range) could be attributed to the protective estrogen effect during this age. In fact, according to current knowledge, women but not men have age-dependent changes in the lipid pattern related to hormonal status,30 and cardiovascular risk increases only after menopause.31,32 In both genders, in the age ranges between 26 and 79 years, we observed an association between high HDL levels and place of residence in the Center of Italy. The same observation, but limited to men in the age range 26–60 years, was seen with BMI. In other words, as the majority of OR values are positive, these associations suggest that the population living in the Center of Italy with ages between 26 and 79 years has a higher risk of having lower HDL levels than the Northern population. In the same age ranges (just failing to reach significance in men aged 61–79 years) and independently from place of residence, obese individuals display a higher risk of having lower HDL levels. With respect to the South, the same positive correlation between place of residence and high HDL was observed in men aged over 41 years, and in women under 40 and over 80 years of age. Interestingly, all associations found between place of residence and high HDL were positive with respect to the North. LDLs also showed a different trend between the Center and the South of Italy as place of residence, in comparison to the North. Thus, our data suggest that in Italy,34 as in other countries,28,29,33 there is a geographical gradient that affects cardiovascular risk, which is not yet completely understood and/or investigated. The associations between LDL levels and place of residence for the oldest subjects generally differed from those observed in the age ranges 61–79 years. This may suggest that after 80 years of age, the lipid parameters are not affected by environmental differences such as the area of residence, and that any possible differences recorded at younger ages are no longer evident, probably because survival for people over 80 is more dependent on family/ genetic factors than on those related to environment/lifestyle acting earlier in the elderly. These data support the results of

NAPOLITANO ET AL.

a recent study on Italian nonagenarians, whose lipid parameters were largely normal, falling within the standard ranges valid for the adult population.35 On the whole, many of the observed associations suggest that geographical differences affect the analyzed parameters, except for subjects older than 80 years, although the results of this study require further experimental validation. In fact, this prototype is a data-sharing study and, as such, it can only identify associations, but cannot provide insight on causation. In conclusion, the HIBP pilot experience demonstrates the feasibility of this type of collaboration and provides a methodological prototype for national epidemiological studies based on the existing data. Furthermore, the consistency of our findings with current knowledge suggests the accuracy of the approach and the validity of the associations found, including some that have not been previously reported. Thus, this study also represents the starting point for further comprehensive analyses that may include other phenotype variables, with the final aim of identifying the determinants of health status in each age group and gender, taking into account areas of residence and sociodemographic factors.

Acknowledgments We acknowledge the collaboration of Isabel Fortier and p3g for their suggestions and helpful assistance. We are thankful to Prof. K.M. Botham for revision of the English language of the manuscript.

Author Disclosure Statement The authors reported no conflicts of interest or financial disclosures. The study was supported by the National Center for Disease Prevention and Control (Ref. ISS IM65).

References 1. Hagen HE, Carlstedt-Duke J. Building global networks for human diseases: Genes and populations. Nat Med 2004;10: 665–667. 2. Fortier I, Doiron D, Little J, et al. International Harmonization Initiative. Is rigorous retrospective harmonization possible? Application of the DataSHaPER approach across 53 large studies. Int J Epidemiol 2011;40:1314–1328. 3. Italian Hub of population biobanks. Available online at: www.iss.it/hibp/. Last accessed 30 June 2014. 4. Public Population Project in Genomics. Available online at: www.p3g.org. Last accessed 30 June 2014. 5. Palmieri L, Panico S, Vanuzzo D, et al. Evaluation of the global cardiovascular absolute risk: The Progetto CUORE individual score. An. Ist Super Sanita` 2004;40:393–399. 6. Tocci G, Ferrucci A, Guida P, et al. Global cardiovascular risk management in different ltalian regions: An analysis of the evaluation of final feasible effect of control training and ultra sensitisation (EFFECTUS) educational program. Nutr Metab Cardiovasc Dis 2012;22:635–642. 7. Mascalzoni D, De Grandi A, Bravo E, et al. Consent harmonization for the Italian hub of population biobank. Biopreserv Biobank 2012;10:A36. 8. Pattaro C, Marroni F, Riegler A, et al. The genetic study of three population microisolates in South Tyrol (MICROS): Study design and epidemiological perspectives. BMC Med Genet 2007;8:29.

HIBP: DATA SHARING PILOT PROJECT

9. Franceschi C, Bezrukov V, Blanche´ H, et al. Genetics of healthy aging in Europe: The EU-integrated project GEHA (GEnetics of Healthy Aging). Ann NY Acad Sci 2007; 1100:21–45. 10. Giampaoli S, Menotti A, Amici R, et al. Prevalenza di alcuni fattori di rischio cardiovascolare in un campione di soggetti anziani. Cardiologia 1992;37:865–870. 11. Giampaoli S, Poce A, Sciarra F, et al. Change in cardiovascular risk factor during 10 years of a community intervention programme. Acta Cardiol 1997;LII(5):411–422. 12. Giampaoli S, Menotti A, Morisi G, et al. Distribuzione di alcuni fattori di rischio cardiovascolare nell’Area Latina Monitoraggio Malattie Cardiovascolari. Rivista Cardiol Preventiva Riabilitativa 1987;5:15–22. 13. Giampaoli S, Vanuzzo D. Atlante italiano delle malattie cardiovascolari. Ital Heart J 2003;4:S1–S121. 14. Brescianini S, Fagnani C, Toccaceli V, et al. An update on the Italian Twin Register: Advances in cohort recruitment, project building and network development. Twin Res Hum Genet 2013;16:190–196. 15. Scafato E, Gandin C, Farchi G. et al. Italian PRoject on the Epidemiology of Alzheimer’s disease (I.PR.E.A.): Study design and methodology of the cross sectional survey. Aging Clin Exp Res 2005;17:29–34. 16. Iacoviello L, Bonanni A, Costanzo S, et al. The Moli-sani Project, a randomized, prospective cohort study in the Molise region in Italy; Design, rationale and objectives. It J Public Health 2007;4:110–114. 17. Italia in cifre 2011. Available from http://www.istat.it/it/ files/2011/06/italiaincifre2011.pdf. Last accessed 30 June 2014. 18. ISTAT. L’Italia in 150 anni. Sommario di statistiche storiche 1861–2010. 2011; 7 339–398 http://www3.istat.it/dati/ catalogo/20120118_00/cap_7.pdf. Last accessed 7 March 2014. 19. Global database on Body Mass Index-BMI Classification. Available from http://apps.who.int/bmi/index.jsp?introPage= intro_3.html. Last accessed 7 March 2014. 20. Report of the National Cholesterol education Program Expert Panel on detection evaluation and treatment of high blood cholesterol in adults. Arch Intern Med 1988;148:36–69. 21. Bachorik PS, Ross JW. National Cholesterol Education Program recommendations for measurement of low-density lipoprotein cholesterol: Executive summary. Clin Chem 1995; 41:1414–1420. 22. Third report of the National Cholesterol education Program Expert Panel on detection evaluation and treatment of high blood cholesterol in adults. Executive Summary. National cholesterol education program. National Heart, Lung and Blood Institute. National Institute of Health. NIH Publication 2001;01–3670. 23. Fraunhofer IBTM Report: Case Study on the Economic Impact of Biobanks Illustrated by EuroCryo Saar, 2009 Available from http://www.tissuebank.it/publicazioni/docUfficiale/

233

24. 25.

26. 27. 28.

29.

30. 31. 32. 33. 34.

35.

DocumentazioneScientifica/BBMRI-Fraunhofer-Case-Studyfinal.pdf. Last accessed 30 June 2014. Williams PH, Schepp K, McGrath B, et al. The stewardship model: Current viability for genetic biobank practice development. ANS Adv Nurs Sci 2010;33:E41–49. Burton PR, Fortier I, Knoppers BM. The Global Emergence of Epidemiological Biobanks: Opportunities and challenges. Building the evidence for using genetic information to improve health and prevent disease. In: Khoury MJ, ed; Human Genome Epidemiology, Oxford University Press, New York 2010;577–599. Bravo E, Napolitano M, Santoro F, et al. The Italian hub of population biobanks as a potential tool for improving public health stewardship. Biopreserv Biobank 2013;11:173–175. BioSHaRE. Available online at: www.bioshare.eu. Last accessed 30 June 2014. Wamala SP, Wolk A, Schenck-Gustafsson K, et al. Lipid profile and socioeconomic status in healthy middle aged women in Sweden. J Epidemiol Community Health. 1997; 51:400–407. Baldassarre D, Nyysso¨nen K, Rauramaa R, et al. Crosssectional analysis of baseline data to identify the major determinants of carotid intima-media thickness in a European population: The IMPROVE study. Eur Heart J 2010; 31:614–622. Haffner SM, Valdez RA. Endogenous sex hormones: Impact on lipids, lipoproteins, and insulin. Am J Med 1995;16: 40S–47S. Barret-Connor E, Bush TL. Estrogen and coronary heart disease in women. JAMA 1991;265:1861–1867. Manson JE. Postmenopausal hormone therapy and atherosclerotic disease. Am Heart J 1994;128:1337–1343. Denarie´ N, Simon A, Chironi G, et al. Difference in carotid artery wall structure between Swedish and French men at low and high coronary risk. Stroke 2001;32:1775–1779. Avogaro A, Giorda C, Maggini M, et al. Incidence of coronary heart disease in type 2 diabetic men and women: Impact of microvascular complications, treatment, and geographic location. Diabetes Care 2007;30:1241–1247. Cevenini E, Cotichini R, Stazi MA, et al. Health status and 6 years survival of 552 90 + Italian sib-ships recruited within the EU Project GEHA (GEnetics of Healthy Ageing). Age (Dordr) 2014;36:949–966.

Address correspondence to: Dr. Elena Bravo Department of Hematology, Oncology and Molecular Medicine Istituto Superiore di Sanita` Viale Regina Elena, 299 00161, Rome Italy E-mail: [email protected]

Development of a pilot project on data sharing among partners of the Italian Hub of Population Biobanks (HIBP): association between lipid profile and socio-demographic variables.

The Italian Hub of Population Biobanks (HIBP) includes both ongoing and completed studies that are heterogeneous in both their purpose and in the spec...
171KB Sizes 0 Downloads 3 Views